error propagation algorithm

The present study investigates the influence of three different microstructure features including volume fraction of α phase (A), thickness of α phase (B), and aspect ratio of primary α (C) on tensile properties of Ti-6Al-4V alloy, by response surface methodology with central composite design (CCD). The experimental data required for the design of experiment (DOE) and analysis of variance (ANOVA) is predicted using the artificial neural network (ANN). First using the experimental data of other researchers, the ANN with two hidden layers by the error propagation algorithm was trained. The main objective of this study is to compare the two feedforward and feedback neural networks in as well as examine the influence of microstructure on the mechanical properties of the Ti-6Al-4V alloy. The results showed that the feedback neural network has higher accuracy than the feedforward neural network to predict the values of yield strength and elongation. Besides, according to ANOVA and response surface method, C, B2, AB2, and A2C factors and A, C, B2, BC, and A2B factors have more significant effects on yield strength and elongation in Ti-6Al-4V alloy, respectively.